AI Paper Atlas

Research map / Detection and segmentation / CNN features

Learning Where To Learn in Cross-View Self-Supervised Learning

Lang Huang, Shan You, Mingkai Zheng, Fei Wang, Chen Qian, Toshihiko Yamasaki

CVPR 2022 · Detection and segmentation › CNN features · approach: Domain-specific segmentation and detection · 44 citations (OpenAlex)

Check your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add LEWEL alignment learning to BYOLBYOL · ImageNet-1K linear classification, 100 epochsAcc@1: 1.3% pointsImproved
Add LEWEL alignment learning to BYOLBYOL · ImageNet-1K semi-supervised classification, 1% labels, 200 pre-training epochsAcc@1: up to 1.3% pointsImproved
Add LEWEL alignment learning to MoCov2MoCov2 · Pascal VOC 07+12 detection, 100 epochsAP: 0.4% pointsImproved
Add LEWEL alignment learning to BYOLBYOL · Pascal VOC 12 semantic segmentation, 100 epochsmIoU: 0.7% pointsImproved
Add aligned loss to global loss (ablation)LEWEL_M · ImageNet-100 linear classificationAcc.: 79.5 -> 81.0Improved
Add aligned loss to global loss (ablation)LEWEL_M · Pascal VOC 12 segmentationmIoU: 61.6 -> 62.7Improved
Couple global and aligned projections through weight sharing (ablation)LEWEL_M with global and aligned losses · ImageNet-100 linear classificationAcc.: 81.0 -> 82.1Improved
Couple global and aligned projections through weight sharing (ablation)LEWEL_M with global and aligned losses · Pascal VOC 12 segmentationmIoU: 62.7 -> 63.4Improved
Replace learned alignment maps with random maps (ablation)LEWEL_M · ImageNet-100 linear classificationAcc.: 82.1 -> 79.8Hurt
Replace learned alignment maps with random maps (ablation)LEWEL_M · Pascal VOC 12 segmentationmIoU: 63.4 -> 61.9Hurt
Increase MoCov2 projection dimension from 128 to 512 (ablation)MoCov2 · ImageNet-100 linear classificationAcc.: 79.5 -> 80.2Improved
Increase MoCov2 projection dimension from 128 to 512 (ablation)MoCov2 · Pascal VOC 12 segmentationmIoU: 61.6 -> 61.5No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy